# 大模型怎么实现连续对话（记忆上下文）

1、OpenAI或者其它大语言模型本身是没有记忆的，如果你不告诉他你之前说了什么以及他之前回答了什么，那么他只会根据你最近一次发送的内容进行回答。

2、所以，要想实现“连续对话”，每次发送消息时，你需要将你之前发送的内容(\*\*user\*\*)以及大模型之前返回的内容(\*\*assistant\*\*)，再结合你本次想发送的内容(\*\*user\*\*) 按 \*\*时序\*\* 组合成一个 messages\[] 数组，然后再将这个数组发送给OpenAI就行了，就是这么简单。

3、有一点需要注意，这样虽然可以实现“连续对话”，但势必造成每次发送的消息内容会非常多，而OpenAI之流是按字数计费的，所以请自行权衡每次应该携带的数量。

官方参考文档：<https://platform.openai.com/docs/guides/text-generation/chat-completions-api>

\> Chat models take a series of messages as input, and return a model-generated message as output.

\> 聊天模型将一系列消息作为输入，并返回模型生成的消息作为输出。

下面是一个实现连续聊天的代码示例，主要看这个 messges 数组：

```python
import openai
from openai import OpenAI
API_SECRET_KEY = "你的智增增获取的api_key";
BASE_URL = "https://flag.smarttrot.com/v1/"; #智增增的base_url
# 连续对话的实例，自己根据业务逻辑填充messages即可
def lianxu():
    client = OpenAI(api_key=API_SECRET_KEY, base_url=BASE_URL)
    resp = client.chat.completions.create(
        model="gpt-3.5-turbo",
        messages=[
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": "Who won the world series in 2020?"},
            {"role": "assistant", "content": "The Los Angeles Dodgers won the World Series in 2020."},
            {"role": "user", "content": "Where was it played?"}
        ]
    )
    print(resp)
```


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.zhizengzeng.com/wen-zhang-lie-biao/da-mo-xing-zen-me-shi-xian-lian-xu-dui-hua-ji-yi-shang-xia-wen.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.